Hi

Key things to remember with clustered indexes:

a) favours range, order, group by sql
b) is the only type of index that results in the data being defraged with
the dbcc index defrag commands
c) is naturally sorted
d) one per table

item c) "may" result in a hot spot if the clustered key is an identity,
causing some blocking and longer internal wait times, therefore reducing the
throughput of your inserts.

If the keys are not identities, you can still have issues with page
splitting, resulting in even more wait time.

These issues are generally answered with some concurreny testing via your
app.

How may rows are you talking about here? do you have any idea of the amount
of data and the predicted % of associated DML activity once you go live?

Cheers


Ck



Oracle LazyDBA home page